Plasma Exosomes Derived from Patients with Primary Immune Thrombocytopenia Attenuate TBX21⁺ Regulatory T Cell-Mediated Immune Suppression via MiR-363-3p

Abstract

Primary Immune Thrombocytopenia (ITP) is characterized by reduced immunosuppressive function of regulatory T cells (Tregs), contributing to immune imbalance and decreased platelet counts. However, the mechanisms behind this reduced efficacy of Tregs remain unclear. Our study used a variety of methods, including Treg function assays, cytokine analysis, and single-cell sequencing, to explore these mechanisms. We found that exosomes from ITP patients inhibited TBX21 expression in Tregs, and impaired their ability to suppress Th1 cells. At the single-cell level, Tregs with high TBX21 expression were identified, and the activity of the TBX21 regulon was found to be enhanced in early-stage Treg subpopulations. We also discovered that ARID3A interacted with SPI1 and TBX21 gene regions, indicating a regulatory relationship between ARID3A, SPI1, and TBX21. Additionally, exosomes in ITP patients' plasma contained elevated levels of miR-363-3p, which negatively correlated with platelet count. These exosomes transferred miR-363-3p to Tregs, downregulating ARID3A, SPI1, and TBX21 expression, thereby weakening Tregs' ability to suppress conventional CD4 + T cells. In conclusion, exosomes from ITP patients reduced Treg function through the ARID3A/SPI1/TBX21 axis by miR-363-3p, diminishing their ability to regulate Th1 cells and contributing to the immune dysfunction observed in ITP.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10753-025-02275-8.

Introduction

Primary Immune Thrombocytopenia (ITP) is an autoimmune hematological disorder, characterized by a decrease in platelet count (< 100 × 10⁹/L) [1, 2]. Clinically, it presents with bleeding in skin and mucosal tissues, and in severe cases, can lead to intracranial hemorrhage [1–3]. Accompanying complications include hypertension, thrombosis, osteoporosis, and others, imposing significant economic burdens on patients’ families and society [1–3]. The clinical treatments are limited, involving therapies such as intravenous immunoglobulin (IVIg) and corticosteroids [1–3]. In addition, several novel therapies have significantly advanced the management of ITP, including thrombopoietin receptor agonists, Rituxan, and Syk and BTK inhibitors [1–7]. These therapies have fundamentally transformed ITP treatment. However, it is crucial to highlight that the precise pathogenesis of ITP remains poorly understood, leading to a lack of targeted therapeutic options. Investigating the immune regulatory mechanisms in ITP patients is of significant clinical value and may provide new avenues for therapeutic intervention.

Regulatory T cells (Tregs) are immune cells with immunosuppressive functions [8]. Forkhead box P3 (FOXP3) is a key transcription factor in Tregs, maintaining their immunosuppressive function and immune tolerance [9–11]. CD25, the α-chain of the interleukin (IL)−2 receptor, helps Tregs receive IL-2 signals, supporting their survival and proliferation [10, 11]. CD127, the α-chain of the IL-7 receptor, is typically low in Tregs, but high in other CD4⁺ T cells [12, 13]. Thus, Tregs are often defined as CD4⁺CD25⁺FOXP3⁺ or CD4⁺CD25⁺CD127^low T cell subset [8–11]. Since FOXP3 is nuclear, we use CD25 and CD127 to sort Tregs to avoid cell damage. In the context of ITP, Tregs play a crucial role in modulating the autoimmune response against platelets and megakaryocytes [14, 15]. Researches showed that Treg exhibit reduced proportions and impaired functions in patients with ITP [14, 16, 17]. Pharmacological strategies aimed at increasing Treg numbers and enhancing their immunosuppressive abilities offer a promising approach to re-establish immune tolerance and, to some extent, restore platelet counts [18–20]. In vivo experiments reveal that mice with Treg deficiencies exhibit increased anti-platelet autoantibodies and a significant reduction in platelet counts, mirroring the pathogenesis seen in human ITP [21].

T-box transcription factor 21 (TBX21)⁺Treg cells (i.e., Treg cells expressing T-bet) represent a distinct subset of regulatory T cells, which primarily function through the regulation of the T-bet transcription factor [22]. The expression of T-bet promotes the enhancement of Treg-mediated immunosuppressive functions [22–24]. TBX21⁺Treg cells play a crucial role in maintaining immune tolerance, preventing autoimmune diseases, and facilitating transplant immune tolerance [22–25]. In autoimmune diseases, the function of TBX21⁺Treg cells may be impaired, leading to the loss of immune tolerance [22–25]. Th1 cells promote immune responses by secreting cytokines such as interferon gamma (IFN-γ) and are involved in antibacterial immunity [26, 27]. However, when Th1 cell activity becomes excessive, it can lead to autoimmune responses and tissue damage [9, 28]. In this context, TBX21⁺Treg cells suppress Th1 cell proliferation and immune activity through direct contact, the secretion of immunosuppressive cytokines, and consumption of growth factors, thereby preventing an overactive immune response [22–25, 29]. Consequently, TBX21⁺Treg cells not only play a key role in maintaining immune tolerance but also protect the body from immune dysregulation by modulating Th1 cell activity [22–25, 29]. This underscores the critical role of TBX21⁺Treg cells in autoimmune diseases.

Exosomes, extracellular vesicles ranging from 30–150 nm in diameter, carry essential proteins, microRNA(miRNA), and metabolic byproducts from their originating cells [30, 31]. CD9 and CD63 are key surface markers of exosomes, widely used for their identification and purification, as they play essential roles in exosome formation, biological function, and distinguishing exosomes from other small vesicles [30–32]. In ITP, plasma exosomes play a role in modulating platelet production by reducing the apoptosis rate of megakaryocytes and limiting platelet production [33]. Exosomes from mesenchymal stem cells can induce regulatory B cell production in ITP patients [34]. However, the precise regulatory mechanisms of plasma exosomes on Tregs within ITP are not yet clear. This study, by probing the role of exosomes in modulating the immune suppressive function of Tregs in ITP patients, seeks to enhance our understanding of autoimmune diseases. The anticipated outcomes promise to unlock new pathways for diagnosing and treating ITP, and to provide a new basis for clinical liquid biopsies.

Materials and Methods

Patient Information and Data Processing

This study involved 15 ITP patients and 15 healthy controls. Data was sourced from the Suzhou 100 Hospital, China. ITP diagnosis adhered to established guidelines. The patient cohort had an average age of 46 years, comprising 7 males and 8 females. The control group consisted of 8 males and 7 females, averaging 45 years in age. In the ITP patient cohort, 7 patients were positive for GPIIb, 8 were positive for GPIIIa, 8 were positive for GPIb/IX, and 7 were positive for GPIb. Four patients had received glucocorticoid treatment, and one patients had been treated with IVIg. Ten patients had a disease duration of less than one month, two patients had a disease duration of more than six months, and three patients had a disease duration between one and three months. The processing and analysis of samples from both groups occurred simultaneously. This research received ethical approval from the ethics committee of Suzhou 100 Hospital, and all participants provided written informed consent upon enrollment.

The single-cell RNA sequencing (scRNA-seq) data of peripheral blood Tregs from healthy individuals were sourced from GSE175602. Monocle analysis was according to protocol on http://cole-trapnell-lab.github.io/monocle-release/, detailed description could be found in our previous research published in 2024 [35]. Single-cell regulatory network inference and clustering by Python (pySCENIC) analysis was according to https://pyscenic.readthedocs.io/en/latest/ [36, 37], the specific workflow was illustrated in Figure S1. Initially, we preprocessed the scRNA-seq data, which included read alignment, quality control, and identification of characteristic genes for each subpopulation. Subsequently, we employed the GEne Network Inference with Ensemble of trees 3 (GENIE3) tool to infer gene regulatory networks based on gene co-expression, identifying candidate regulons. Next, using the cisTarget tool, we filtered out genes not directly regulated by transcription factors based on their regulatory motifs, thereby establishing regulons comprising transcription factors and their directly regulated genes. Following this, we calculated the activity of each regulon at the single-cell level using the AUCell method and set thresholds to determine whether cells were regulated by the respective regulons. Finally, we visualized the results.

Chromatin immunoprecipitation sequencing (ChIP-seq) data were from GSE106034, GSE246060, GSE43119 [38–40]. The process began by using sratoolkit to download raw data files from the Sequence Read Archive website (https://www.ncbi.nlm.nih.gov/sra) and converting them into fastq format. Quality control was performed. After assessing the quality, low-quality sequences were removed and adaptors were trimmed. The cleaned sequences were aligned to the hg38 reference genome, and the resulting sam files were converted to bam format. These bam files were then visualized using the Integrative Genomics Viewer browser.

Flow Cytometry Analysis and Treg Sorting

For surface staining, peripheral blood samples were collected from the healthy and ITP patient using venipuncture, and peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation using Ficoll-Paque. The PBMCs were then prepared into a single-cell suspension. Red blood cells were lysed using Red Blood Cell Lysis Buffer, incubated for 30 min at 4 °C. The cells were washed twice with PBS containing 2% fetal bovine serum (FBS) to remove any residual lysing reagent. After washing, the cells were resuspended to a concentration of 10⁶ cells/mL in PBS. Approximately 0.1 mL of the suspension was taken out, blocked using Fc block, for 30 min at 4 °C, and flow cytometry antibodies were added. The cells were incubated with antibodies for 30 min at 4 °C in the dark, then washed twice with PBS containing 2% FBS to remove excess antibodies. Finally, the cells were resuspended in PBS and analyzed by flow cytometry.

For intracellular staining, surface staining was first performed as described above. Following surface staining, cells were permeabilized using a membrane permeabilization buffer. A working solution was prepared at a ratio of 1:3 of permeabilization concentrate to diluent. Each tube received 500–1000 μL of the working solution and was incubated at room temperature for 0.5–1 h in the dark. After permeabilization, about 2 mL of diluted buffer was added to each tube for washing twice, and finally, the cells were resuspended in 0.1 mL buffer, blocked. The staining step was carried out at room temperature for 30 min, then the cells were resuspended, and detected.

For Treg sorting, surface staining was carried out as above. Since staining for FOXP3 required cell permeabilization, flow cytometric gates were set to sort out alive CD8⁻CD4⁺CD25⁺CD127^low Tregs and CD8⁻CD4⁺CD25⁻ conventional CD4⁺T cells. The gating strategy for Treg and conventional CD4⁺T cells, along with purity assessment post-sorting, was shown in Figure S2. For samples analyzed by flow cytometry, we typically used the gating strategy of CD8⁻CD4⁺CD25⁺FOXP3⁺ to define Tregs. The detailed information of antibodies was shown in Table S1.

Treg Suppression Experiments and the Examination of Cytokines

For the Treg suppression assay, conventional CD4⁺T cells were labeled with CFSE and served as responders. These CFSE-labeled responder cells were co-cultured with Tregs at varying ratios (Treg:responder = 1:1 to 1:8) in 96-well plates pre-coated with 2 μg/mL anti-CD3 and 1 μg/mL anti-CD28. After 96 h of co-culture, the cell proliferation was assessed by flow cytometry based on CFSE dilution. Control groups were responder cells cultured without Tregs. Flow cytometric analysis was performed using a BD LSRFORTESSA X-20 flow cytometer (BD Biosciences). The data were analyzed using FlowJo software.

For cytokine detection, cells were stimulated for 4 h with 2 μg/mL anti-CD3 and 2.5 μg/mL anti-CD28. During the final two hours of stimulation, 1 μL/mL GolgiStop was added to block cytokine secretion. After stimulation, cells were fixed and permeabilized using the BD Cytofix/Cytoperm™ kit, following the manufacturer's protocol. Cells were then stained with antibodies against cytokines such as IFN-γ using a BD cytokine staining kit. The full list of antibodies and other reagents used is provided in Table S1. Cytokine expression was analyzed by flow cytometry using a BD LSRFORTESSA X-20 flow cytometer (BD Biosciences).

Exosome Isolation and Analysis

Blood was collected from the venous arm of the subjects into EDTA anticoagulant tubes, then centrifuged at approximately 2500 rpm for 10–15 min, and the upper plasma layer was carefully collected. Plasma samples were collected and resuspended in PBS (pH 7.4) to a final volume of 10 mL. The samples were first centrifuged at 800 g for 5 min to remove any cell debris, and then at 2,000 g for 10 min to further clarify the supernatant. The resulting supernatant was carefully collected and filtered through a 0.22 μm filter. The filtered supernatant was then ultracentrifuged at 120,000 g for 4 h at 4 °C to pellet the exosomes. The pellet was resuspended in PBS and subjected to a second round of ultracentrifugation at 120,000 g for 4 h. This process was repeated twice to ensure high purity of exosomes. After the final ultracentrifugation, the exosome pellet was resuspended in PBS for further analysis.

The isolated exosomes were diluted 50-fold in PBS and analyzed using a Nanosight NS300 instrument (Malvern Instruments). The analysis was repeated five times to obtain an average size distribution and the absolute number of exosomes. The size distribution was determined by the Nanosight software, and the concentration was calculated based on the particle count.

For AllMag® Superparamagnetic Silica Nanobeads-based flow cytometry detection of exosome surface molecules, a 10 mM 2-(N-morpholino)ethanesulfonic acid (MES) solution (containing 0.05% Tween, MEST) was prepared and adjusted to pH 6.0 using 0.01 M NaOH solution, then stored at 4 °C. 2 mg of AllMag® Superparamagnetic Silica Nanobeads were added to 500 µL of MEST solution and washed twice, followed by magnetic separation to remove the supernatant. Next, the magnetic beads were activated by preparing N-(3-Dimethylaminopropyl)-N-ethylcarbodiimide hydrochloride Crystalline (EDC) (5 mg/mL) and N-Hydroxysuccinimide (NHS) (5 mg/mL) solutions, mixing them with the beads, and incubating at 37 °C for 30 min.

For antibody coupling, 50 µg of antibody was added to the activated beads, and the mixture was incubated at 37 °C for 30 min, followed by magnetic separation, PBS washing twice, and PBS (containing 0.05% Tween, PBST) washing once. Exosome staining was performed using 30 µM DIO dye to label the exosome membrane (incubated for 30 min), followed by magnetic separation and resuspension in PBS for flow cytometry analysis. The excitation wavelength was 484 nm, and the emission wavelength was 501 nm. For antibody labeling, the Alexa Fluor® 488 Microscale Protein Labeling Kit was used to label anti-CD9/CD63 antibodies. First, 1 mL of 1 M sodium bicarbonate solution was prepared, and 100 µL of 500 µg/mL antibody solution was added, mixed well. 10 µL of dH2O was added to dissolve Alexa Fluor® 488 TFP ester (concentration 11.3 nmol/µL). The required volume of fluorescent dye was calculated based on the protein concentration and molecular weight, and added to the antibody solution. After a 15-min incubation at room temperature, free unbound dye was removed using resin purification. The labeled antibody solution was then mixed with 50 µL of exosome solution (6.0 × 10⁹ particles/mL) and incubated at room temperature for 30 min. Flow NanoAnalyzer was used for analysis, and the data were processed using FLOW JO 7.6 software. The information of antibodies and other reagents was shown in Table S1.

Transfection Experiment

After passaging 293 T cells, four plasmids, including two hsa-miR-363-3p inhibitor (5’-UACAGAUGGAUACCGUGCAAUU-3’) and two corresponding vector (5’-CAGUACUUUUGUGUAGUACAA-3’), both sourced from Sangon Biotech, were co-transfected into the cells using lipofection. The cells were cultured in vitro for 24 h, and the fluorescence signals from the plasmids were observed under a fluorescence microscope. We selected the plasmids with higher transfection efficiency and their corresponding transfected cells for subsequent experiments. After 48 h of culture, the supernatant was collected by ultracentrifugation to concentrate the lentiviral solution. The concentrated virus was then used to infect 293 T cells, and the infection efficiency was assessed 48–72 h post-infection. If the efficiency exceeded 70%, the infection was considered highly efficient. Infected cells were passaged 1–3 times, and the stability of the infection was confirmed by the continued observation of green fluorescence. The same procedure was applied for transfecting the target sequence into Tregs.

Real-Time Fluorescent Quantitative PCR

Total RNA was extracted using Trizol, and the RNA was reverse transcribed into cDNA, ensuring an OD260/OD280 > 1.9. DNA template, primers, Oligo(dT)18, reaction mix, enzyme mix, and DEPC water were added for detection. The relative expression level of the target gene was calculated.

Statistical Analysis

For the comparison between two groups of continuous variables, if the assumptions of homogeneity of variances and normal distribution were met, t-test was used; otherwise, non-parametric test was applied. Comparison between paired samples was conducted using paired t-test or non-parametric test. Data were presented as means ± standard deviation. The correlation analysis between two variables was conducted using the Spearman method. The significance level was set at 0.05.

Results

The Plasma Exosomes of ITP Patients Weakened the Ability of Treg to Suppress CD4⁺IFN-γ⁺Cells

We collected and identified exosomes from plasma samples. Utilizing the Nanosight instrument, we determined that the concentration of the exosome solution was 1.71 × 10⁸ ± 2.00 × 10⁷ particles/mL, their average diameter was 87.25 nm (Fig. 1A). The CD9⁺exosomes accounted for 84.4% and 77.4% in the healthy group and ITP group, respectively, while CD63⁺ exosomes accounted for 82.3% and 84.6%, respectively (Fig. 1B).

Fig. 1.

Plasma exosomes from ITP patients weakened the ability of Treg to inhibit the Th1 (CD4⁺IFN-γ⁺) cell subset. (A) The Nanosight instrument was used to measure the size distribution and counts of exosomes, with results repeated five times. (B) CD9 and CD63 markers on exosomes were measured by flow cytometry. (C) Plot the proportion of CD25⁺FOXP3⁺Treg cells within the CD8⁻CD4⁺T cell gate and compare the their proportions between ITP patients and healthy controls. (D) Flow cytometric analysis results and statistical chart of the Treg inhibition function experiment. (E) The changes in Treg inhibition ability on Th1/Th2/Th17 in ITP after the addition of pExo. Exo, exosomes; pExo, exosomes derived from ITP patients; *P < 0.05; ***P < 0.001; ns, no significance

We discovered a notable decrease in the proportion of Tregs in the peripheral blood of ITP patients compared to that in healthy subjects, a difference that was statistically significant (P < 0.001, Fig. 1C). Then, we isolated Tregs from both ITP patients and healthy individuals. In subsequent experiments involving CFSE-labeled conventional CD4⁺T cells, we conducted tests to assess the suppressive function of Tregs. The findings were revealing: compared to the control group of healthy individuals, the Tregs from ITP patients demonstrated a diminished capacity to inhibit the proliferation of conventional CD4⁺T cells when co-cultured with exosomes from ITP patients (pExo), a result that was statistically significant (P < 0.05, Fig. 1D). However, the proliferation of conventional CD4⁺T cells remained largely unchanged when pExo was co-cultured without Tregs (Fig. 1D). This suggests that plasma exosomes in ITP patients primarily impact the suppressive function of Tregs, leading to reduced inhibition of conventional CD4⁺T cell proliferation.

To delve deeper into the influence of pExo on the Tregs’ ability to suppress various conventional CD4⁺T cell subgroups in ITP patients, we analyzed the cytokines in conventional CD4⁺T cells across the designed groups as shown in Fig. 1E. It was observed that the introduction of pExo significantly weakened the Tregs’ capacity to suppress the Th1 (CD4⁺IFN-γ⁺) subgroup in ITP(Fig. 1E). In contrast, the suppression of Th2 and Th17 cells by Tregs did not exhibit any statistically significant alterations (Fig. 1E). Moreover, after the addition of pExo, the suppressive ability of Tregs from healthy individuals on Th1 cells remained unchanged (Figure S3).

ITP patient plasma exosomes inhibited TBX21 in Tregs, with TBX21 highly expressed in C14 Treg subset and high TBX21⁽⁺⁾ regulon activity in C13 subgroup.

Based on literature reports, TBX21⁺ Tregs can suppress Th1 cells [29]. Our observations indicated that after interaction with pExo, the transcription level of TBX21 in Tregs from ITP patients decreased (Fig. 2A), potentially explaining why pExo reduces the ability of Tregs to suppress Th1 cells.

Fig. 2.

The impact of plasma exosomes from ITP patients on TBX21 levels in Tregs and the expression patterns of TBX21 in different Treg subsets. (A) Comparison of mRNA levels of various transcription factors emphasized in Tregs between ITP Tregs and ITP Tregs after co-culture with pExo. (B) Uniform Manifold Approximation and Projection (UMAP) visualization of Treg scRNA-seq data, TBX21 expression in different subgroups and the enrichment results of genes highly expressed in C14. (C) Heatmap and bubble chart displaying the relative activity, regulon specificity scores, and expression levels of different regulons. (D) Top four or five regulons with high specificity scores in C13 and C14. pExo, exosomes derived from ITP patients; GENIE3, GEne Network Inference with Ensemble of trees 3; C, cluster; TF, transcript factor; pct.1, target genes’ expression level in C14; pct.2, target genes’ expression level in other subgroups except C14; AUC, area under the curve. ***P < 0.0001; ns, no significance

Analysis of scRNA-seq data of human Tregs [41] revealed that the Cluster (C)14 subpopulation exhibited high TBX21 expression (Fig. 2B). Genes highly expressed in the C14 subpopulation were primarily enriched in pathways related to Immune system processes, Positive regulation of T cell activation, Regulation of lymphocyte proliferation, and Regulation of memory T cell differentiation (Fig. 2B). This indicates that C14 represents a specialized subset of Tregs with proliferative and differentiative capabilities.

To further investigate the gene regulatory networks characteristic of Treg subpopulations, we employed the pySCENIC [36, 37] method to analyze the scRNA-seq data, identifying the transcription factor regulatory networks within Tregs (Figure S1). Our analysis identified distinct transcription factor regulatory clusters across different Treg subpopulations (Fig. 2C). TBX21⁽⁺⁾ regulon activity was elevated in the C13 subset, and the expression level of genes within the TBX21⁽⁺⁾ regulon was also increased in C13 (Fig. 2C-D). In the C14 subset, AT-rich interaction domain 3A (ARID3A)⁽⁺⁾ regulon activity, regulatory specificity, and expression levels were significantly elevated(Fig. 2C). ARID3A⁽⁺⁾ regulon activity and regulatory specificity were also increased in the C13 subset, though less pronounced than in C14 (Fig. 2C). Spi-1 proto-oncogene (SPI1)⁽⁺⁾ regulon specificity scores were elevated in both C13 and C14 (Fig. 2D).

Given the high expression of TBX21 in C14, we hypothesize that the increased activity of the TBX21⁽⁺⁾ regulatory cluster in C13 may endow C13 with the potential to express TBX21. By increasing TBX21 expression, the C13 subset could differentiate into C14, ultimately forming a C14 subpopulation with stable high TBX21 expression. However, we acknowledge that to conclusively demonstrate the gradual increase and eventual stabilization of TBX21 expression during the differentiation of C13 to C14, further experimental validation is required.

C13 differentiated into C14, during which the ARID3A⁽⁺⁾ and SPI1⁽⁺⁾ regulons were highlighted.

Using Monocle pseudotime analysis, we observed that subclusters C13 and C14 were in the early stages of Treg differentiation (Fig. 3A). This suggests that these two cell populations might serve as precursor cells for Treg self-renewal. We then extracted C13 and C14 for cell differentiation trajectory analysis, which revealed five distinct differentiation states (Fig. 3B). C13 was in the relatively early differentiation stage and could differentiate into C14 (Fig. 3B).

Fig. 3.

Differentiation of C13 into C14 highlighting ARID3A⁽⁺⁾ and SPI1⁽⁺⁾ regulons. (A) Pseudotime trajectories of Tregs and their various subpopulations. (B) Pseudotime analysis results for C13 and C14. (C) Target genes, NES, Q-value, and motif logo of regulons with high specificity in either C13 or C14, along with their enrichment analysis results. (D) Heatmap depicting the transcript levels of genes within regulons listed in Fig. 3C, showing changes along the pseudotime. (E) Target genes exhibiting changes due to pseudotime and different differentiation states. C, cluster; NES, normalized enrichment score

According to Fig. 2D, the principal regulons in C13 and C14 included ARID3A⁽⁺⁾ and SPI1⁽⁺⁾. In C13, the top four specific regulons also included TBX21⁽⁺⁾ and KLF transcription factor 4 (KLF4)⁽⁺⁾, whereas in C14, the top five specific regulons included MAF bZIP transcription factor B (MAFB)⁽⁺⁾ and JunD proto-oncogene (JUND)⁽⁺⁾. The corresponding motifs and target genes of these transcription factors were shown in Fig. 3C. These regulons exhibited significantly high regulatory activity in Tregs (Q-value < 0.05). Gene enrichment analysis on these regulons indicated that they were primarily enriched in pathways related to T cell differentiation, T cell activation, Response to stress, Immune response regulation, and Immune effector processes (Fig. 3C). This suggests that these highly specific regulons in C13 and C14 can influence Treg differentiation, activation, proliferation, and promote Treg immune response under certain stress conditions.

Subsequently, we analyzed the expression changes of genes within these specific regulons in C13 and C14 over the differentiation timeline. We observed that genes within ARID3A⁽⁺⁾, SPI1⁽⁺⁾, KLF4⁽⁺⁾, and MAFB⁽⁺⁾ exhibited overall increased expression in the terminally differentiated C14 cluster (Fig. 3D). TBX21⁽⁺⁾ was highly expressed in the C14 cluster, consistent with earlier results, while natural killer cell granule protein 7 (NKG7) was highly expressed in the C13 cluster (Fig. 3D-E), which might have explained the overall higher expression of TBX21⁽⁺⁾ in C13.

Further analysis revealed that S100 calcium binding protein A8 (S100A8) in ARID3A⁽⁺⁾ and major histocompatibility complex, class II, DP alpha 1 (HLA-DPA1), major histocompatibility complex, class II, DR beta 1 (HLA-DRB1), cathepsin S (CTSS), and lysozyme (LYZ) in SPI1⁽⁺⁾ initially increased in the precursor C13 cells before branching point 2, then decreased in the terminal C13 cells after branching point 2, and subsequently increased in the terminal C14 cells after branching point 2 (Fig. 3E). Additionally, under the regulation of SPI1⁽⁺⁾, KIAA0513 and myeloid cell nuclear differentiation antigen (MNDA) showed significantly increased expression in C14 during differentiation (Fig. 3E). These indicate that under the regulation of ARID3A⁽⁺⁾ and SPI1⁽⁺⁾, these genes have the potential for expression in the precursor C13 cells and exhibit higher, more stable expression as C13 differentiated into C14.

In ITP patients, exosomal miR-363-3p downregulated ARID3A in Treg, resulting in decreased TBX21 levels and consequently weakening Treg immunosuppressive function.

To further identify which regulons in Tregs played a crucial role in ITP patients, we examined the key transcription factors identified in our initial screening by qRT-PCR. We observed a significant reduction in SPI1, ARID3A, and TBX21 in Tregs from ITP patients after the addition of pExo compared to ITP patient Tregs alone (Fig. 4A).

Fig. 4.

Plasma exosomal miR-363-3p from ITP patients downregulated ARID3A in Tregs, subsequently reducing TBX21 expression and weakening the immunosuppressive function of Tregs. (A) Transcript levels of key transcription factors in Tregs from ITP patients before and after the addition of pExo. (B) We analyzed the ChIP-seq data for ARID3A and SPI1 to investigate their binding interactions with TBX21, ARID3A, and SPI1. Promoter and enhancer regions of these genes were annotated using the UCSC Genome Browser (https://genome.ucsc.edu/). Additionally, transcription factor binding sites (TFBS) were identified and annotated using the JASPAR database. TFBS refers to regions where transcription factors bind, and JASPAR (JAva Tools for Sequence Analysis and Recognition) is a database for predicting and analyzing these binding sites. (C) Comparison of target miRNA levels across different groups. (D) Alignment sites between the ARID3A 3’UTR and miR-363-3p, along with results from Treg suppressive function experiments and miR-363-3p levels after rescue experiments. (E) Transcript levels of target genes across different groups and correlation between platelet counts and Exosomal miR-363-3p levels. (F) Illustrative summary of findings from this study. Our study found that exosomal miR-363-3p reduced ARID3A in Tregs, leading to lower levels of SPI1 and TBX21 and finally impairing Tregs’ immunosuppressive function on Th1 cells. We observed the Treg subset with elevated transcriptional regulatory activity of the TBX21⁽⁺⁾ regulon gradually differentiated into the high TBX21-expressing subpopulation, with ARID3A⁽⁺⁾ and SPI1⁽⁺⁾ regulons emphasized during this process. These subclusters, at the initiation stage of Treg differentiation, functioned as precursor cells and maintained Treg self-renewal. This might explain why exosomes in ITP patients, leading to a reduction in TBX21⁺Tregs, could impair their immunosuppressive capability. inhi, inhibitor; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no significance; r, correlation coefficient obtained using Spearman method

Through the analysis of ChIP-seq data for ARID3A and SPI1, we identified binding interactions between ARID3A and both TBX21 and SPI1, as well as between SPI1 and TBX21 (Fig. 4B). We further annotated the promoter and enhancer regions of these genes using the UCSC Genome Browser (https://genome.ucsc.edu/) (Fig. 4B). Using established chromatin marks, H3K27Ac and H3K4Me3, we determined that regions with elevated peak intensities close to the transcription start site (TSS) (within 1000 bp) correspond to promoter regions, while regions further from the TSS (greater than 1000 bp) are classified as enhancers. We found that the peak with the strongest binding of ARID3A to TBX21 was not located near either the promoter or enhancer regions, whereas the peak with the strongest binding of SPI1 to TBX21 was situated within both the promoter and enhancer regions. This suggests that SPI1 may have a more direct regulatory effect on TBX21 (Fig. 4B). Furthermore, the peaks where ARID3A and SPI1 interacted with ARID3A were located within the enhancer region of ARID3A (Fig. 4B), and the peaks where ARID3A and SPI1 interacted with SPI1 were located within the enhancer region of SPI1 (Fig. 4B). These findings suggest that ARID3A's regulation of SPI1 is more direct, while its regulation of TBX21 is less direct when compared to SPI1. Additionally, we also identified interactions between ARID3A and S100A8, as well as between SPI1 and CTSS, KIAA0513, and MNDA (Figure S4).

Previous studies have shown that exosomal microRNAs can influence the expression of various proteins in target cells [42, 43]. To determine whether plasma exosomes from ITP patients regulated TBX21 by modulating ARID3A expression in Tregs, we examined several microRNAs potentially binding to ARID3A. We found that only miR-363-3p was significantly elevated in Tregs from ITP patients after the addition of pExo compared to ITP patient Tregs alone (Fig. 4C). Furthermore, miR-363-3p levels were higher in plasma exosomes from ITP patients compared to healthy controls (Fig. 4C), consistent with previous reports [44].

We identified multiple binding sites for miR-363-3p within the 3’ Untranslated Region (UTR) region of ARID3A, encompassing both the seed region and the 3’ pairing region of miR-363-3p (Fig. 4D). The seed region of miR-363-3p corresponded to the 7mer-m8 sequence in the 3’ UTR of ARID3A (Fig. 4D). To further investigate the regulatory effects of miR-363-3p on related molecules and functions in Tregs, we co-cultured Tregs with pExo and then added a miR-363-3p inhibitor or control sequence. We assessed the immunosuppressive function of Tregs. The results showed that Tregs had a reduced ability to inhibit conventional CD4⁺T cells after co-culturing with pExo; however, this inhibition was restored when miR-363-3p was inhibited (Fig. 4D). We also found that the transcription levels of ARID3A, TBX21, and SPI1 decreased in Tregs after adding pExo, but were restored with the miR-363-3p inhibitor (Fig. 4E). Further studies revealed a negative correlation between miR-363-3p levels in exosomes and patient platelet counts (Fig. 4E).

As shown in Fig. 4F, these findings suggest that plasma exosomes in ITP carried miR-363-3p, which bound to the 3’ UTR of ARID3A in Tregs, leading to mRNA degradation and reduced expression. This reduction in ARID3A might decrease its binding to SPI1, downregulating SPI1 and its downstream pathways, ultimately affecting TBX21. Additionally, ARID3A could directly influence TBX21, and its downregulation led to decreased TBX21 levels, resulting in a reduction of TBX21⁺Tregs. According to the literature, TBX21⁺Tregs inhibit Th1 cells [23, 29]. Therefore, after co-culturing Tregs with pExo in ITP patients, the ability to inhibit Th1 cells was diminished.

Discussion

ITP is characterized as an autoimmune hematologic disorder [3]. Patients, suffering from reduced platelet counts, manifest symptoms ranging from skin and mucosal bleeding to hemorrhaging in vital organs [2, 3, 16]. The treatments for ITP are primarily pathogenesis-based [4–6]. Both the latest international consensus and ASH guidelines emphasize that the treatment of ITP should be tailored to the individual, considering clinical presentation, disease course, and response to different therapies [4–6].

We found that the proportion of Tregs was reduced in ITP (vs the healthy, Fig. 1C). Studies have shown that a reduction in Tregs is a key factor in the pathogenesis of ITP, leading to dysregulated immune responses [45–48]. In mouse models of ITP, Tregs levels were significantly reduced, while B-cell depletion therapy increased their proportions [21, 49]. Tregs play a crucial role in restoring immune tolerance and balancing T cell subsets in ITP, as their enhancement by low-dose decitabine alleviates thrombocytopenia and suppresses pro-inflammatory Th1 and Th17 cells [18]. Tregs play a pivotal role in ITP by promoting immune tolerance and inhibiting the phagocytosis of antibody-coated platelets by macrophages, with their restoration in function and numbers significantly ameliorating the pathological progression of ITP [19]. Exploring immune regulatory mechanisms in ITP patients offers valuable insights for identifying effective therapeutic targets.

Our findings revealed a critical role of exosomes in the plasma of ITP patients. These exosomes attenuated the ability of Tregs to inhibit the proliferation of conventional CD4⁺T cells, as evidenced by Fig. 1D. CD4⁺T cells, including Th1 (CD4⁺IFN-γ⁺), Th2 (CD4⁺ IL-4⁺), and Th17 (CD4⁺ IL-17A)⁺) subgroups, are found to be notably increased in patients with recurrent and refractory ITP [18, 50]. Pharmacologically inhibiting Th1 elevation significantly alleviates clinical symptoms in ITP patients [18, 50]. We found that posting co-culture with the patients’ plasma exosomes, Tregs demonstrated a significantly reduced suppression of Th1 cells, whereas their inhibitory effect on Th2 and Th17 subgroups remained largely unaltered (Fig. 1E). This leads us to hypothesize that exosomes might contribute to ITP pathogenesis by impairing Treg-mediated suppression of Th1 cells.

In our study, we observed that the transcription level of TBX21 in Tregs from ITP patients decreased following the introduction of pExo (Fig. 2A). However, when Tregs from ITP patients interacted with exosomes from healthy donors, and when Tregs from healthy donors interacted with exosomes from ITP patients, no significant change in TBX21 levels was observed in Tregs (Figure S5). This suggests that exosomes derived from ITP patients carry specific genetic factors that can affect Tregs under the pathological state of ITP, leading to TBX21 dysfunction in Tregs. Studies have demonstrated that TBX21⁺Tregs selectively accumulate in ovarian tumor sites to modulate T cell responses, thereby limiting effective antitumor immune responses [51]. Activation of TBX21 enhances the capacity of FOXP3⁺Tregs to counterbalance Th1-type inflammation [52]. The loss of function or elimination of TBX21-expressing Tregs but not the mere loss of TBX21 expression within Tregs results in severe Th1 autoimmunity [23]. Conversely, depletion of TBX21⁻Tregs leaves the remaining TBX21⁺ cells to specifically inhibit Th1 responses [23]. In the C14 subset, the high expression of TBX21 further regulates its specific immune functions. These findings provide new insights into the complex role of TBX21 in immune regulation.

Through further investigation, we found that TBX21 is highly expressed in the C14 subset of Tregs, while the activity of the TBX21⁽⁺⁾ regulon is specifically enhanced in C13 (Fig. 2). To further explore the relationship between C13 and C14, we performed trajectory analysis, which revealed that C13 has the potential to differentiate into C14 (Fig. 3). As shown in Table S2, the features of the C13 population include genes associated with T cell development, proliferation, and differentiation [53–57], such as RPS17, RPS29, TRAC, and RPL36A, whereas C14 expresses more genes related to immune functions, antigen presentation, and immune response [58], such as TBX21 and HLA-DRA [59–61]. The expression changes of these genes may serve as markers for the differentiation of C13 into C14, suggesting that C13 may transition to C14 through the activation of specific transcription factors and signaling pathways. Since C13 and C14 are newly defined Treg subsets, no literature has yet reported Treg populations with these two cell characteristics or their differentiation relationship. We found that ARID3A⁽⁺⁾ and SPI1⁽⁺⁾ regulons are differentially expressed between the C13 and C14 groups (Fig. 3D). As C13 gradually differentiates into C14, the transcription levels of the ARID3A⁽⁺⁾ and SPI1⁽⁺⁾ regulons progressively increase (Fig. 3D). In contrast, during the self-renewal process of C13 (from C13 to C13), the transcription levels of the ARID3A⁽⁺⁾ and SPI1⁽⁺⁾ regulons remain unchanged or decrease (Fig. 3D-E), indicating that these regulons may play a role in the differentiation process from C13 to C14. However, whether they directly influence this differentiation process requires further confirmation through knockdown experiments. Although previous studies have reported upregulation of ARID3A expression in liver macrophages and peripheral monocytes in primary biliary cholangitis and primary sclerosing cholangitis [62], the correlation between ARID3A expression and increased disease severity in systemic lupus erythematosus suggests that ARID3A or its downstream targets could serve as novel therapeutic targets for systemic lupus erythematosus [63]. SPI1 as a key transcription factor, regulates the progression of ankylosing spondylitis through NF-κB signaling by modulating TLR5 [64]. Enhanced expression/phosphorylation of JAK2/STAT5A and increased SPI1 expression were observed in the lesions of giant cell arteritis patients [65]. TBX21 + Tregs accumulate in Th1 cell-mediated inflammation, where TBX21 expression is crucial for their homeostasis and function during type 1 inflammation [66]. After ChIP-seq analysis, we found that ARID3A binds to the enhancer region of SPI1, while SPI1 binds to both the promoter and enhancer regions of TBX21 (Fig. 4B). However, the interactions between ARID3A, SPI1, and TBX21 have not been reported, which represents the innovative aspect of this study.

Moreover, our results indicated that in ITP patients, exosome-carried miRNA-363-3p increased (Fig. 4C). As we known, mRNAs could be degraded after interacting with miRNA on 3’UTR region [67, 68]. Studies have demonstrated that exosomal miRNA-208b can promote Treg expansion in colorectal cancer [69]. In multiple sclerosis patients, the exosomal miRNA profile differs from that of healthy controls, with a significant increase in miRNA-let-7i, which inhibits Treg cell induction by targeting insulin-like growth factor 1 receptor and transforming growth factor-beta receptor 1 [70]. Integrating our findings with existing literature above, we hypothesize that miR-363-3p carried by plasma exosomes in ITP patients may target the 3’UTR of ARID3A in Tregs, leading to the degradation of its mRNA. The downregulation of ARID3A further results in decreased levels of SPI1 and TBX21, ultimately reducing the number of TBX21⁺Tregs. This reduction weakens the suppressive capability of Tregs on Th1 cells in ITP patients.

In summary, as shown in Fig. 4F, we discovered that in ITP patients, Tregs receiving exosomes exhibited reduced ARID3A levels due to exosome-carried miRNA-363-3p. This reduction might lead to decreased SPI1 levels, ultimately downregulating TBX21 in Tregs and impairing their immunosuppressive function on Th1 cells. Additionally, we identified a Treg subpopulation with high TBX21⁽⁺⁾ regulon activity that further differentiated into a subgroup with elevated TBX21 expression under the influence of ARID3A⁽⁺⁾ and SPI1⁽⁺⁾ regulons. These two cell groups were at the initial stages of Treg differentiation and crucial for maintaining Treg self-renewal. This explained why reduced TBX21 in Tregs of ITP patients weakened their immunosuppressive capability.

However, this study had certain limitations. While we provided data on the in vitro suppression of conventional CD4⁺T cells by Treg cells, these findings were limited by the inability to replicate in vivo conditions. Specifically, the in vitro experiments could not mimic the process by which Tregs migrate to areas of Th1 or CD8 + T cell accumulation in response to chemotactic signals [23, 52], a process that might have influenced immune regulation. Furthermore, the interactions between Tregs and conventional CD4⁺T cells in vitro were solely mediated through direct contact and cytokine secretion, whereas in vivo, these interactions are governed by a more complex immune microenvironment. Therefore, future studies should incorporate animal models and more precise techniques to further validate the role of T-bet in Tregs within the in vivo context.

Although our study provides valuable insights into the role of exosomes in modulating Tregs in ITP patients, the specific involvement of the ARID3A/SPI1/TBX21 pathway in this process remains unclear. As illustrated in Fig. 4F, the mechanistic details linking this pathway to exosome-mediated Treg regulation have yet to be fully elucidated. Further investigations are needed to confirm whether this pathway plays a pivotal role in the modulation of Tregs by exosomes in the context of ITP. Future studies focusing on the mechanistic interactions between ARID3A, SPI1, and TBX21, as well as their potential modulation by exosome signaling, will be critical to advancing our understanding of Treg dysfunction in ITP. Additionally, exploring in vivo models and patient-specific data could help to validate these findings and clarify the clinical relevance of the ARID3A/SPI1/TBX21 pathway in ITP pathogenesis.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ARID3A

AT-rich interaction domain 3A

AUC

Area under the curve

C

Cluster

ChIP-seq

Chromatin immunoprecipitation sequencing

CTSS

Cathepsin S

Exo

Exosomes

FOXP3

Forkhead box P3

GENIE3

GEne Network Inference with Ensemble of trees 3

HLA-DPA1

Major histocompatibility complex, class II, DP alpha 1

FBS

Fetal bovine serum

HLA-DRB1

Major histocompatibility complex, class II, DR beta 1

EDC

N-(3-Dimethylaminopropyl)-N-ethylcarbodiimide hydrochloride Crystalline

IFN-γ

Interferon gamma

IL

Interleukin

Inhi

Inhibitor

ITP

Primary Immune Thrombocytopenia

IVIg

Intravenous immunoglobulin

JUND

JunD proto-oncogene

KLF4

KLF transcription factor 4

LYZ

Lysozyme

MAFB

MAF bZIP transcription factor B

miRNA

MicroRNA

MES

2-(N-morpholino)ethanesulfonic acid

MNDA

Myeloid cell nuclear differentiation antigen

NES

Normalized enrichment score

NHS

N-Hydroxysuccinimide

NKG7

Natural killer cell granule protein 7

ns

No significance

PBMCs

Peripheral blood mononuclear cells

PBST

PBS containing 0.05% Tween

pct.1

Target genes’ expression level in C14

pct.2

Target genes’ expression level in other subgroups except C14

pExo

Exosomes from ITP patients

pySCENIC

Single-cell regulatory network inference and clustering by Python

r

Correlation coefficient obtained using Spearman method; S100A8, S100 calcium binding protein A8

scRNA-seq

Single-cell RNA sequencing

SLE

Systemic lupus erythematosus

SPI1

Spi-1 proto-oncogene

TBX21

T-box transcription factor 21

TF

Transcript factor

Tregs

Regulatory T cells

TSS

Transcription start site

UMAP

Uniform Manifold Approximation and Projection

UTR

Untranslated Region

Authors’ Contributions

Tengda Li, wrote the R codes, analyzed the data, wrote the paper; Tengda Li, Ying Xu, Peng Liu and Yuanlan Huang, did the experiment; Cheng Qian and Tianqin Wu, provided the clinical samples; Tengda Li, edited the paper; Peng Liu, Yuanlan Huang and Tengda Li, revised the manuscript. The authors have accepted responsibility for the entire content of this manuscript and approved its submission.

Funding

This research was supported by Wenzhou Basic Public Welfare Research Project No.Y2023091/2023Y0784, Ministry of Education Industry-University Cooperation Talent Cultivation Program (240805635262430), Zhejiang Provincial Natural Science Foundation of China under Grant No. LQ24H100005, Scientific research business expenses of Wenzhou Medical University No.KYYW202318, Talent Research Startup Project of Wenzhou Medical University No.QTJ21020, Zhejiang Province College Student Science and Technology Innovation Program (New Talent Program, 2024R413A004), Project ZR2021MH239 supported by Shandong Provincial Natural Science Foundation. This research was supported by Biomedical Big Data Intelligent Computing Center of Oujiang Lab.

Data Availability

No datasets were generated or analysed during the current study.

Declarations

Ethics Approval and Consent to Participate

All participants were given written informed consent. The results were assessed blindly for all the experiments using clinical samples to ensure scientific rigor. All procedures in this study were approved by the ethics committee of Suzhou 100 Hospital on December 31, 2022 (clinical trial number 2022KY201).

Conflict of Interest

The authors declare no competing interests.